Amazon Nova助力功夫源：提升金融数据分析效率，推动量化投资普惠化

import boto3
import json


client = boto3.client("bedrock-runtime",region_name = 'us-east-1')
MODEL_IDs = ["us.amazon.nova-pro-v1:0","us.amazon.nova-lite-v1:0","us.amazon.nova-micro-v1:0"]
model = MODEL_IDs[0]
def nova(model,prompt):
    lm_system = """ You are young, critical of reality, thoughtful, witty in language, quantitative trading expert and with a keen sensitivity and rigor towards data. Please answer business questions related to quantitative trading according to the requirements and strictly follow the way of reply by requirement. Your thinking of chain should follow the steps in <amazon_nova_thinking_protocol> tags.


<amazon_nova_thinking_protocol>
...
</amazon_nova_thinking_protocol>
"""
    prompt = prompt
    messages = [
        {
            "role": "user", 
            "content": [{"text": prompt}]
        },
        {
            "role": "assistant", 
            "content": [{"text": "Here is my JSON analysis result:\n```json"}]
        }
    ]
    # stopSequences for end with }
    inf_params = {"maxTokens": 300, "topP": 0.1, "temperature": 0.1, "stopSequences": ["}"]}


    additionalModelRequestFields = {
        "inferenceConfig": {
            "topK": 20
        }
    }
    try:
        model_response = bedrock.converse(
            modelId=model_id, 
            messages=messages, 
            system=system, 
            inferenceConfig=inf_params,
            additionalModelRequestFields=additionalModelRequestFields
        )


        response_body = model_response["output"]["message"]["content"][0]["text"]
        return response_body
    
    except Exception as e:
        print(f"Error invoking Bedrock: {str(e)}")
        return None

Your task is to analyze the content within <origin_input> tag and extract valuable insights related to opportunities and risks and give a professional Analysis Report (sometimes output as JSON). Remember "Q&As" stands for "Questions and Answers".


Take a little breath and analyse the content within <origin_input> tag step by step.


Ensure that the analysis covers every single trading day within the specified period, with a focus on unusual movements, significant changes in technical indicators, and overall market trends.


<Market_Analysis_Protocol>
  <Analysis_Process>
    # Analysis Steps:
      1. Comprehensive Daily Analysis:
        - Analyze EVERY SINGLE TRADING DAY in the input data, from the earliest to the latest date.
        - For each date, compare its data to the average of the *previous* 5 trading days.
        - Identify unusual movements based on the defined criteria for EACH DAY.
        - Ensure equal depth of analysis for all dates, avoiding bias towards recent data.


      2. Historical Context Review:
        - Identify patterns, trends, and unusual movements across the entire date range.
        - Focus on major trends and recurring patterns relevant to the entire period.


      3. Comprehensive Evaluation:
        - Analyze factor interaction patterns and behavioral shifts across the entire period.
        - Identify overarching trends and significant changes in market structure.
  
    # Signal Validation Steps:
      1. Individual Factor Analysis
        - Compare to thresholds
        - Check time consistency
        - Verify signal strength
      
      2. Cross-Factor Confirmation
        - Minimum 2-factor alignment
        - Time window matching
        - Pattern consistency


      3. False Signal Filtering
        - Threshold validation
        - Pattern completion check
        - Multi-timeframe confirmation
        
    # Output Compilation:
      - Ensure ALL identified unusual movements for each trading day, are grouped under the same trading day entry, and included in the final output.
      - Maintain chronological order in reporting unusual movements.
      - Provide a balanced analysis that gives equal weight to all trading days in the dataset.
  </Analysis_Process>


  <Unusual_Movement_Detection>
    ...
  </Unusual_Movement_Detection>


  <Data_Analysis_Requirements>
    ...
  </Data_Analysis_Requirements>


  <Output_Standards>
    ...
  </Output_Standards>


  <Data_Structure>
    ...
  </Data_Structure>


  <Critical_Reminders>
    ...
  </Critical_Reminders>


</Market_Analysis_Protocol>

#####
【思考过程】


嗯，这是一个关于东风股份（600006）的多维度市场分析。我需要仔细梳理每个交易日的异常变动，特别关注最近几天的趋势和潜在影响。


首先，我注意到几个关键的异常变动时间点：
1. 20241223和20241224是最近的两个交易日，异常变动较为集中
2. 20241217、20241218、20241219也出现了一些显著的异常指标


我将逐步分析这些异常变动，重点关注：
- 指数涨幅异常
- 技术指标（KDJ、乖离率）变化
- 个股与行业相对涨幅
- 市场结构和资金流向


对于20241223和20241224，异常变动较为复杂：
- 指数涨幅出现异常
- KDJ指标的K、D值发生显著变化
- 个股和行业涨幅出现较大偏离
- 乖离率持续处于负值且变化明显


这些异常变动背后可能隐含着市场情绪和资金流向的重要信号。我需要深入解析每一个异常点的具体含义和潜在影响。


关键分析维度：
1. 技术面：KDJ指标显示持续低迷
2. 相对涨幅：个股表现显著弱于大盘和行业
3. 趋势判断：短期内呈现震荡下行态势


接下来我将系统性地梳理每个异常变动点的具体implications。

{
    "all_unusual_movements": [
        {
            "trading_day": "20241224",
            "name": "多维度异常",
            "descriptions": [
                {
                    "name": "指数涨幅",
                    "description": "较5日平均增长1.318%，高于历史波动范围"
                },
                {
                    "name": "行业涨幅",
                    "description": "较5日平均增长1.605%，显著高于大盘表现"
                },
                {
                    "name": "D值",
                    "description": "出现异常波动，显著低于前期水平"
                }
            ],
            "analysis": "市场出现短期反弹迹象，行业板块表现强于大盘，但个股未能充分受益。技术指标显示市场情绪仍处于谨慎状态，可能面临震荡调整压力。",
            "value_judgement": "negative"
        },
        {
            "trading_day": "20241223",
            ...
        },
        {
            "trading_day": "20241217",
            ...
        }
    ],
    "overall_analysis": [
        {
            "name": "技术面",
            "content": "近期股价持续承压，KDJ指标显示市场情绪低迷。未来3-5天可能面临震荡调整，但存在技术性反弹可能。",
            "value_judgement": "negative"
        },
        {
            "name": "资金面",
            ...
        }
    ],
    "risks": [
        {
            "name": "技术性调整风险",
            "description": "股价持续低于10日均线，可能触发进一步下探。建议密切关注支撑位情况。"
        },
        {
            "name": "行业分化风险",
            "description": "个股表现显著落后于行业平均水平，可能面临持续分化压力。"
        }
    ],
    "opportunities": [
        {
            "name": "技术性反弹机会",
            "description": "当前股价已严重偏离均线，短期内存在技术性反弹可能。建议关注低吸时机。"
        },
        {
            "name": "行业轮动可能",
            "description": "行业板块整体表现相对强势，个股调整可能为新的投资机会提供契机。"
        }
    ]
}